Good morning. Today marks our last report of 2024. We’ll be taking some time to rest and recharge for what promises to be a busy, busy 2025 (and we hope you are all doing the same).

We’ll see you in the new year, and in the meantime, hope you all enjoy a week of relaxation, comfort food and family time. Happy holidays!

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🤖 AI for Good: Wearable robots

• 👁️‍🗨️ Biden launches investigation into Chinese chips

• 🏛️ Europe’s AI Code of Practice nears publication

• ⚡️ Report: The data center energy spike

AI for Good: Wearable robots

Researchers at the Exoskeleton Laboratory at the Korea Advanced Institute of Science and Technology (KAIST) recently developed a lightweight, wearable exoskeleton — inspired by “Iron Man” — to help paralyzed people walk again. 

The details: Unlike other exoskeletons, the WalkON Suit F1 was designed to autonomously walk up to and attach to its user, even if the user is sitting in a wheelchair. This enables a paraplegic person to get into the suit without additional help. 

• The suit comes loaded up with sensors and cameras, in addition to advanced motion control and computer vision algorithms, which work together to help the robot predict movement, retain balance and avoid obstacles. 

• Competing recently in the Cyborg Olympics, an international competition focused on evaluating robotic assistance for disabled people, the KAIST team took home a gold medal as the only team capable of completing all the required tasks. 

Why it matters: For years now, researchers have been examining ways in which machine learning can be leveraged to produce advanced prosthetics, not just for paraplegics, but also for people who have lost limbs. That promise is now coming into reach. 

Biden launches investigation into Chinese chips

President Joe Biden’s administration said Monday that it is launching an investigation into foundational, or legacy, chips made in China. The legacy chips that are the target of the inquiry are not as advanced as the chips used for artificial intelligence applications, though they are used in a number of non-AI applications. 

The details: The U.S. Trade Representative, in the months-long probe that will follow, will first assess whether China’s actions are “unreasonable” and burdensome.

• This, as Bloomberg pointed out, opens up the impending Trump administration to impose tariffs on the sector. 

• The current administration said it is still exploring ways to shore up and centralize the semiconductor supply chain. 

The landscape: This follows a recent round of export restrictions from both the U.S. and China that was focused on restricting the flow of materials necessary to build advanced computer chips. 

The President-elect, meanwhile, has promised to impose tariffs of as much as 60% on Chinese goods. 

• The latest report from the National AI Opinion Monitor found that privacy and security remain key concerns around AI: “Americans are particularly concerned about AI’s influence on politics (58%) and news media (53%), likely reflecting fears of misinformation and manipulation during the 2024 election cycle.”

• Last week, a bipartisan US House of Representatives Task Force on Artificial Intelligence published a massive report detailing recommendations on AI to Congress. Kate Brennan, associate director of the AI Now Institute, said the report gestures “towards the many material harms of rapid AI adoption while providing few meaningful policy recommendations to address it.”

• Palantir and Anduril join forces with tech groups to bid for Pentagon contracts (FT).

• AI is here to stay, so we may as well learn to use it? That’s a cop-out (Times Higher Education).

• OpenAI has edge over Google in winning publishers’ business (The Information).

• Exclusive — Microsoft works to add non-OpenAI models into 365 Copilot products, sources say (Reuters).

• The Philippines is creating a ChatGPT rival that speaks Filipino and Taglish (Rest of World).

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

Europe’s AI Code of Practice nears publication

Last week, the second draft of the European Commission’s general purpose AI Code of Practice was published, bringing the commission one step closer to eventually releasing a finalized version of the Code. 

The details: The Code is designed as a guiding source for ensuring developers remain compliant with the EU’s AI Act, which began entering into force earlier this year. 

• The draft thus far details copyright and transparency obligations for developers of general-purpose generative AI systems, which will include summaries of the content used to train each model. 

• The second part of the draft Code applies only to developers of the “most advanced” generative AI models, and includes a list of safety and transparency obligations. 

The group plans to publish the third draft of the Code of Practice in February of next year; the intention is to publish a final version of the Code by May of 2025. 

The landscape: Europe remains one of few countries committed to regulating AI, with regulatory efforts stalled out or stagnant in the U.S. and the U.K., despite significant public support for legislation that would rein the technology in. 

Report: The data center energy spike 

A recent report from the U.S. Department of Energy (DOE) found that data center electricity load growth has tripled over the past 10 years, and is expected to double again between 2024 and 2028. 

The details: The report attributes the growth in U.S. electricity demand to the “rise of artificial intelligence” and the steady expansion of the data centers used to power the tech. 

• According to the report, data centers consumed 4.4% of total U.S. electricity in 2023 and are expected to consume as much as 12% of total electricity demand by 2028. 

• In 2014, data centers used 58 terawatt hours (TWh) of electricity; in 2023, they used 176 TWh and in 2028, they are expected to use as much as 580 TWh. 

For context, Greece, Switzerland and Israel each consumed around 300 TWh of electricity in 2023. 

“The United States has seen an incredible investment in artificial intelligence and other breakthrough technologies over the last decade and a half, and this industrial renaissance has created greater demand on our domestic energy supply,” U.S. Energy Secretary Jennifer Granholm said in a statement. “We can meet this growth with clean energy.”

The energy department is pushing both industrial energy storage and on-site energy generation for data center operators as part of its plan to make data centers a “grid asset rather than a burden.”

The department said that the commercialization of technologies including more efficient semiconductors, advanced nuclear energy, long-term energy storage and geothermal energy will all be vital to meet this steadily rising demand. 

Geothermal, according to the International Energy Agency, currently meets less than 1% of the world’s energy demand. But the IEA recently found that investment in geothermal is growing, with the sector expected to draw some $1 trillion globally by 2035. 

As costs continue to fall, the IEA predicted that geothermal has the potential to provide some 15% of the world’s energy demand by 2050. 

This comes as the energy demand of AI has pushed the bulk of the Big Tech sector to pursue the construction of new nuclear power plants. In the meantime, new research has highlighted the massive impact of data center emissions on public health around the world.

When I think about the cost-benefit analysis of AI use and the lack of transparency in the industry, I think about energy use and emissions. All we know is that energy demand keeps spiking, pushing far ahead of innovations in technological efficiency. 

We need to re-ground our collective understanding of chatbots and the internet at large, not as some mystical, on-demand portal, but as the result of extensive hardware, powered by enormous quantities of electricity and emitting enormous quantities of carbon and other particulate matter. 

We need to refocus our attention on achieving innovations in technological efficiency, and we need to think more about the cost of use associated with these systems. 

🤔 Your thought process:

Selected Image 2 (Left):

• “The flames on these candles were more realistic and variable than those of Image 1.”

Selected Image 1 (Right):

• “The light on the second seems artificial.”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you next year.

Here’s your view on o3:

33% of you think it is time for some regulation already.

26% said it should be released; 15% said it shouldn’t.

18% don't think it’s actually that powerful.

Depends:

• “It should NOT be released until there is enough enforceable regulation and a monitoring framework to ensure no one ignores probable risk because of financial gain.”

Not that powerful:

• “They already proved they overhype and underdeliver with Sora.”